University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

Papers in Natural Resources Natural Resources, School of

2009 MOVEMENTS, DISTRIBUTION, AND ABUNDANCE OF GREAT ARGUS (ARGUSIANUS ARGUS) IN A SUMATRAN RAINFOREST Nurul L. Winarni University of Georgia

Timothy G. O'Brien Wildlife Conservation Society, [email protected]

John P. Carroll University of Georgia, [email protected]

Margaret F. Kinnaird Wildlife Conservation Society

Follow this and additional works at: http://digitalcommons.unl.edu/natrespapers Part of the Natural Resources and Conservation Commons, Natural Resources Management and Policy Commons, and the Other Environmental Sciences Commons

Winarni, Nurul L.; O'Brien, Timothy G.; Carroll, John P.; and Kinnaird, Margaret F., "MOVEMENTS, DISTRIBUTION, AND ABUNDANCE OF GREAT ARGUS PHEASANTS (ARGUSIANUS ARGUS) IN A SUMATRAN RAINFOREST" (2009). Papers in Natural Resources. 658. http://digitalcommons.unl.edu/natrespapers/658

This Article is brought to you for free and open access by the Natural Resources, School of at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Papers in Natural Resources by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. The Auk 126(2):341 350, 2009 ‘ The American Ornithologists’ Union, 2009. Printed in USA.

MOVEMENTS, DISTRIBUTION, AND ABUNDANCE OF GREAT ARGUS PHEASANTS (ARGUSIANUS ARGUS) IN A SUMATRAN RAINFOREST

NURUL L. WINARNI,1,2 TIMOTHY G. O’BRIEN,3,4,5 JOHN P. C ARROLL,2 AND MARGARET F. K INNAIRD3,4

1Wildlife Conservation Society, Indonesia Program, Jl. Ciremei 8, P.O. Box 311, Bogor 16003, Indonesia; 2D.B. Warnell School of Forest Resources, University of Georgia, Athens, Georgia 30605, USA; 3Wildlife Conservation Society, 2300 Southern Boulevard, Bronx, New York 10460, USA; and 4Mpala Research Center, P.O. Box 555, Nanyuki, Kenya

Abstract.—We used radiotelemetry, habitat sampling, camera trapping, and line-transect surveys to explore movement patterns, distribution, and abundance of Great Argus Pheasants (Argusianus argus) in Sumatra, Indonesia. We radiotracked six adult and one subadult males. Territories averaged . o . ha, and home-range size did not vary by month or by relative abundance of selected plant foods. Daily travel distance ( o  m) varied significantly between months but did not reflect changes in plant foods. Territories were used almost exclusively by resident males. Males preferentially used undisturbed forest (habitat I). Vegetation structure at male display sites and random points indicated that display sites were located in undisturbed forest, with few lianas and small leaf size on trees adjacent to the display site. Between  and , we conducted five line-transect surveys in conjunction with camera-trap surveys. Density estimates of calling males varied from . to . males km−, and the total density estimate ranged from . to . km−. Density estimates increased substantially between  and , reflecting recovery from depressed densities after the – El Niño drought. Habitat occupancy estimates varied from % to % but were not significantly different over time. The proportion of occupied habitat was similar to the proportion of habitat I. We conclude that Great Argus Pheasants prefer undisturbed forest and rarely use other habitat even as population density increases. Restricted movements and habitat preference may limit the ability of Great Argus Pheasants to colonize forest fragments. Received  October , accepted  November .

Key words: Argusianus argus, density, distribution, Great Argus , habitat preference, male movements.

Movimientos, Distribución y Abundancia de Argusianus aarhus en un Bosque Lluvioso de Sumatra

Resumen.—Usamos radiotelemetría, muestreos de hábitat, captura con cámaras fotográficas y transectos lineales para explorar los patrones de movimiento, la distribución y la abundancia de Argusianus aarhus en Sumatra, Indonesia. Seguimos a seis machos adultos y a un subadulto con radiotelemetría. Los territorios fueron en promedio de . o . ha, sin variación en los ámbitos de hogar entre meses o en relación con la abundancia relativa de algunas especies de plantas que sirven como alimento. Las distancias de movimiento diario ( o  m) variaron significativamente entre meses pero esta variación no se relacionó con los cambios en las plantas que sirven como alimento. Los territorios fueron usados casi exclusivamente por machos residentes. Los machos prefirieron áreas de bosque no intervenido (hábitat I). La estructura de la vegetación en las áreas de exhibición de los machos y en puntos distribuidos al azar indicó que las áreas de exhibición estaban localizadas en bosques sin intervención, con pocas lianas y con árboles de hojas pequeñas en las áreas adyacentes a los sitios de exhibición. Entre  y , realizamos muestreos en cinco transectos lineales y utilizando cámaras fotográficas. Las estimaciones de densidad de machos vocalizando varió entre . y . machos km−, y el estimado de densidad total varió entre . y . aves km−. Las estimaciones de densidad aumentaron considerablemente entre  y , reflejando una recuperación de las densidades bajas que se registraron después de la sequía causada por El Niño entre  y . Las estimaciones de ocupación de hábitat variaron de un % a un % pero éstas no fueron significativamente diferente entre periodos de tiempo. La proporción de hábitat ocupado fue similar a la proporción de hábitat I. Concluimos que estos faisanes prefieren bosques no intervenidos y que rara vez utilizan otros tipos de hábitat, incluso cuando las densidades poblacionales aumentan. El movimiento restringido y las preferencias de hábitat de esta especie pueden limitar su colonización de fragmentos de bosque.

5Address correspondence to this author. E-mail: [email protected]

The Auk, Vol. , Number , pages  . ISSN -, electronic ISSN -. ‘  by The American Ornithologists’ Union. All rights reserved. Please direct all requests for permission to photocopy or reproduce article content through the University of California Press’s Rights and Permissions website, http://www.ucpressjournals. com/reprintInfo.asp. DOI: ./auk.. — 341 — 342 — WINARNI ET AL. — AUK, VOL. 126

The Great Argus Pheasant (Argusianus argus), one of the most loss. Today, % of the original lowland forest area remains, world’s largest pheasants, is restricted to tropical forests of the and much of that is fragmented. Most Sumatran galliforms are Sunda Shelf, including the Malay Peninsula, Sumatra, and Bor- forest interior specialists and are believed to be sensitive to the neo (Smythies , MacKinnon and Phillipps ). The effects of forest fragmentation (McGowan and Gillman , Wi- includes two subspecies, the Bornean Great Argus Pheasant (A. a. narni et al. ). The Great Argus Pheasant is considered “near grayi) and the Malay–Sumatra Great Argus Pheasant (A. a. argus). threatened” by IUCN (Fuller and Garson ), and it is protected The species is sexually dimorphic, with males averaging . kg by law in Indonesia. However, other than records of occurrence and females . kg. Males have iridescent oceli on the large sec- and the surveys by Nijman () and Winarni et al. (), little ondary feathers and elongate central retrices on the tails, which information is available on this threat classification or on the ef- are a. m in length (Delacour , Johnsgard ). The mating fects of habitat disturbance on pheasants. system is polygynous, with males advertising on dancing sites in Here, we report on the ecology of Great Argus Pheasants in an exploded-lek system (Davison b, Johnsgard ). Davison a national park in southern Sumatra. We combine radiotelemetry, (a) reported, on the basis of vocalizations and foraging obser- habitat analysis, occupancy analysis, and population estimation to vations, that both males and females may defend territories. Lit- explore the characteristics and spatial distribution of home ranges tle is known about the timing of reproduction in the wild. Beebe of males and to estimate the density and distribution of this low- reported one recently hatched nest, probably in October (Johns- land-forest population. gard ). Davison (a) discovered a nest in June and reported a breeding season from February to August (Davison b). In METHODS Sumatra, Indonesia, we observed a nest with two eggs in August  (M. F. Kinnaird unpubl. data). Madge and McGowan () Study area.—This research was conducted in Bukit Barisan Sela- concluded that the breeding season was unclear but may occur be- tan National Park (BBSNP), Sumatra (Fig. ), in the southwestern tween March and July. In captivity, females have laid – clutches part of the island (n`−n`S and n`−n`E). The third- of two eggs per year, with clutches in every month (Johnsgard largest protected area in Sumatra (, km; O’Brien and Kin- ), which suggests the possibility of multiple clutches per year naird ), BBSNP contains some of the largest remaining tracts in the wild. The Great Argus Pheasant’s omnivorous diet includes of lowland rainforest in Sumatra and functions as the primary fruits, seeds, flowers, leaf buds, and leaf-litter invertebrates. Beebe watershed for southwest Sumatra (Kinnaird et al. ). It is bor- () reported that the Borneo subspecies consumed primarily dered by villages, agriculture, and plantation forestry and suffers ants, but also other invertebrates, leaves, nuts, fallen fruit, and from encroachment for logging and agriculture, as well as from il- seeds. On Peninsular Malaysia, Davison (a) reported that the legal hunting (Kinniard et al. , O’Brien et al. ). The park Malay–Sumatra subspecies’ diet was rich in ants but dominated covers an altitudinal gradient from sea level to , m, and for- by fruits of the Palmae, Annonaceae, and Fabiaceae. est habitats include lowland, hill, and montane forests. Rainfall is The species inhabits primary forest but can also be found in slightly seasonal, averaging  mm month− (, mm year−) old secondary forest and, rarely, in young secondary forest (Nij- during the present study. Temperatures normally fluctuate be- man ). Despite their size and their loud, distinctive call, a pen- tween n and nC. The region is subject to droughts during El etrating two-note kwow wow, the birds are difficult to observe, and Niño–Southern Oscillation (ENSO) events, when monthly rain- there have been only two systematic studies of the species (Davi- fall may drop to  mm. The present study was initiated in the year son a, Winarni ) and one long-term survey (O’Brien and following an especially severe ENSO drought that included forest Kinnaird ). Great Argus Pheasants appear to prefer lowland fires in part of the study area. and hill forests below , m with a low density of undergrowth The research was conducted in the surroundings of the Way and presence of climbing vines (Davison a, Nijman ). Nij- Canguk Research Station (n`pS, n`pE), in the south- man () found that Great Argus Pheasants on Borneo were less western part of BBSNP (Fig. ). The station is located in lowland common near villages, but on Sumatra, O’Brien et al. () and forest and hosts a high diversity of wildlife, including several en- O’Brien and Kinnaird () found that Great Argus Pheasant dangered large mammals, such as Sumatran Tiger (Panthera abundance was high in forests adjacent to villages and that dis- tigris), Sumatran Elephant (Elephas maximus), Sumatran Rhinoc- tance to forest edge or villages made little difference in the species’ eros (Dicerorhinus sumatrensis), and  species of birds, including occurrence and abundance. The differences between studies may  pheasant species (Winarni ). The study area encompasses a be attributable to higher hunting pressure in Borneo, where Great ,-ha forest block within a larger forest matrix and includes Argus Pheasant feathers are used in traditional dances. Great Ar- a grid of trails at -m intervals. This gridded area is bisected gus Pheasants occur at densities of .–. birds km− (both sexes) by the Canguk River, and the two sub-areas are referred to as in Malaysia and .–. birds km− (calling males) in Borneo (Da- “north side” and “south side.” All trails are permanently marked at vison and Scriven , Nijman ). -m intervals. The study area contains a mosaic of lowland habi- The lowland and hill forests of Indonesia have suffered seri- tat types, including primary forest, lightly disturbed forest, and ous deforestation in recent decades. National average annual de- previously burned forest (D. S. Busch et al. unpubl. data, Sunarto forestation is estimated at . million ha since  (Holmes ). ). The last category resulted from fires during the – The island of Sumatra has experienced the highest rates of defor- and – ENSO droughts (Kinnaird and O’Brien ). estation in Indonesia; between  and , Sumatra lost nearly Radiotelemetry.—Male Great Argus Pheasants were cap- a third of its forest cover (Kinnaird et al. ). Lowland and hill tured using a modification of a traditional leg-snare device, con- forests that support the highest biodiversity have experienced the sisting of a thin platform of twigs attached to a triggered snare, APRIL 2009 — GREAT ARGUS PHEASANT MOVEMENT AND ABUNDANCE — 343

FIG. 1. Location of the Way Canguk study area, Sumatra, Indonesia. which was anchored to a flexible fiberglass fishing-rod tip. As standard measurements and released the near the capture the bird stepped on the platform, the twigs collapsed, triggering site. After capture, to reduce stress on the birds, all snares were the snare, which was tightened by a quick tug of the fishing-rod removed from the site and human activity was reduced in the area tip. Capture attempts were focused on males with active dancing for several days before tracking began. grounds. Typically, males with dancing grounds use two or three Birds were tracked for three months during September– entrances (Davison b, N. Winarni pers. obs.), so we set five November , depending on date of capture. We assume that to seven snares at or near entry points. We also set snares along this was late in the breeding season on the basis of observations the trail system in the vicinity of dancing grounds. Snares were by Davison (b) and because males actively maintained dis- constructed the day before deployment and set before dawn. We play sites throughout the study. Radio signals were monitored us- checked the snares on the basis of the birds’ calling behavior. Dur- ing Wildlife Materials TRX-S (– MHz) receivers and ing preliminary observation, we found that birds usually call from three-element Yagi antennae. Individuals were first located within a branch adjacent to the dancing ground before entering it. We let a -ha block delineated by the trail grid. We used triangulation the bird perform its first morning call before checking snares and by two teams of observers spaced – m apart to collect then checked at two-hour intervals until  hours. pairs of bearings of the individual simultaneously from the grid Each captured bird was fitted with a -MHz necklace-type transect system. Usually, triangulation bearings were taken from radiotransmitter (model A, Advanced Telemetry Systems, distances < m from the bird, which minimized the chance Isanti, Minnesota), leg band, and numbered collar. We collected of signal bounce and produced % error polygons < m 344 — WINARNI ET AL. — AUK, VOL. 126

(White and Garrott ). To determine daily home range and kernel home ranges based on %, %, and % probability of use movements, we conducted intensive daily tracking twice a month (Hooge and Eichenlaub ). Compositional analysis was con- for each bird from  to  hours, with locations taken at -h ducted in two steps, following Aebischer et al. (). We analyzed intervals. In addition to intensive tracking, we obtained a loca- () the proportion of MCP versus study area and () proportions of tion for each bird on  days each month. Home-range size was radiolocations within the home ranges. We used MACOMP.SAS calculated from tracking data using % minimum convex poly- (Ott and Hovey ) to conduct compositional analysis of pooled gon (MCP; White and Garrott ). Kernel home range (White data from the south and north study areas. Zero values, which are and Garrott ) was used to determine utilization distribution unutilized habitat, were replaced by .%. within home ranges and centers of activities based on a % prob- We attempted to locate all dancing grounds throughout the ability of use. We performed repeated-measures analyses of vari- study site. We compared macrohabitat classification of display ance to test whether there is a significant effect of time on home sites with expected values based on the proportion of each habitat ranges, daily travel, and mean distances of radiolocations to danc- category in the study area. Next, we paired active dancing grounds ing ground. All home ranges, movement patterns, and distance of and random sites within delineated home ranges for microhabi- travel were calculated using ARCVIEW, version ., with tat comparisons. On each of the active dancing grounds and ran- Movement extensions (Hooge and Eichenlaub ). dom sites, a -m-radius circular plot was laid out. Within this Food abundance.—Because Great Argus Pheasants con- circular plot, we counted number of trees and measured the DBH sume primarily fruits and other plant materials (Davison a), of five nearest trees with DBH ≥ cm, measured the distance to we attempted to measure availability of fruits, flowers, seeds, and the center of the dancing ground, and recorded the size of five fungi. In the vicinity of each dancing ground, we placed two to leaves for each of these five trees. Size of leaves was divided into sixteen -m transects spaced  m apart. We used the existing three types: small (length a cm, width a cm), medium (length grid trails as the starting points of transects. We attached a -m – cm, width a cm), and large (length  cm, width  stick on a -m measuring tape along each transect, and at -m cm). We also counted the number of fallen logs and the distance intervals we recorded the number of fallen fruits, flowers, seeds, to the center of the dancing ground. At four different bearings, we and mushrooms present beneath the stick. This process was re- measured understory density, canopy openness, and litter thick- peated monthly to develop an index of abundance for plant food ness. Understory density was estimated using coverage of a  r  types. We used multiple regressions to test whether monthly m sheet divided into  r  cm grids. We used a spherical den- home-range size fluctuated as a function of changes in the abun- siometer to measure canopy openness at the center and at four dance index for flowers, fruits, seeds, and mushrooms considered random locations within the plot. We also recorded position of separately or as the summed food abundance index. Similarly, we dancing ground as on or off a human trail, presence or absence tested the effect of plant food abundance index on daily ranging of a game trail, whether the site was on level ground or on a ridge, patterns and distance moved from the dancing ground. All statis- and the relative density score of dominant understory plant types tical analyses were performed using SPSS, version .. (SPSS, (seedlings and saplings, lianas or climbers, grass herbs, gingers). Chicago, Illinois). Scoring of understory plant density was divided into a four-point Habitat use.—Macrohabitat classification followed the struc- scale based on percent coverage of each type of plant within the tural classification of D. S. Busch et al. (unpubl. data), modified to plot: score  (–%), score  (–%), score  (–%), and score incorporate changes attributable to the  ENSO fire (Kinnaird  (–%). and O’Brien , T. O’Brien unpubl. data). Data were collected Camera trapping and line transects.—We conducted camera- in  circular plots of radius  m placed systematically along the trap sampling in the south Way Canguk research area on five oc- trail system throughout the study area. The classification of habi- casions between August  and November  to estimate the tats was based on size distribution, height, and density of trees proportion of occupied habitat in the study area. We used pas- with diameter at breast height [DBH] ≥ cm, sapling density and sive infrared camera traps (CamTrak South, Watkinsville, Geor- size distribution, understory density, and canopy openness. Dis- gia) with data packs that stamp each photograph with time and turbance was indicated by the presence of rattan and other palms date of the event. We deployed  camera traps in a systematic ar- (Palmae), lianas, bamboo (Poaceae), and wild ginger (Zingiber- rangement throughout the study area. Cameras were left in place aceae; Whitten et al. ). The macrohabitats of the study area for ~ days and then retrieved. After films were processed, we were classified at the scale of  ha and divided into four habitat scored each photograph for number of birds in the frame and sex types: (I) undisturbed forest with large trees (.%), (II) undis- of each bird, when possible. Adult Great Argus Pheasants are eas- turbed forest with small trees (.%), (III) disturbed forest with ily sexed, but chicks resemble females until first molt, when male large trees (.%), and (IV) disturbed forest with small trees characteristics become apparent (Madge and McGowan ). (.%). We also scored each photograph as an independent event on the Locations of radiotagged Great Argus Pheasants and % basis of O’Brien et al.’s () criteria to minimize the possibility MCP were overlaid on a habitat classification map and classi- of counting individuals twice and used only independent photo- fied by habitat type using ARCVIEW and the Spatial Analyst ex- graphs for analysis. We scored each day of sampling at each sam- tension. Habitat use by radiotagged Great Argus Pheasants was ple point for presence of pheasants. compared with habitat availability using compositional analysis To estimate the proportion of the study area occupied by (Aebischer et al. ). The MCP home range of each individual Great Argus Pheasants, we used occupancy analysis (MacKenzie was analyzed using ARCVIEW with Animal Movement exten- et al. , ). Eight-day sampling intervals were combined sions (Hooge and Eichenlaub ). In addition, we also analyzed into four sampling replicates, and we used methods described by APRIL 2009 — GREAT ARGUS PHEASANT MOVEMENT AND ABUNDANCE — 345

MacKenzie et al. () to estimate site occupancy (ψ) and detec- predator while in the snare. Eight birds were fitted with radiotrans- tion probability (p) for each survey. We considered three possible mitters, but one of these died soon afterward and was presumed outcomes: () the site is occupied and Great Argus Pheasants are to have been killed by a predator. One adult male’s radio failed in detected, ψ r p; () the site is occupied but Great Argus Pheas- the final month of the study. We analyzed data for six adult males ants are not detected, ψ r ( − p); and () the site is unoccupied, that maintained dancing grounds and one sub-adult male without a  − ψ. In these analyses, we assumed that detection probabilities dancing ground. We recorded – radiolocations bird− (Table ). – and site occupancy were constant across time and space within Total home-range sizes varied, ranging from  to  ha (x  each sample period (model ψ(.)p(.)). Although there are alterna- . o .; Table ). The sub-adult male’s home range was twice as tive multiseason models and covariate models that might better large as that of the largest adult male. Home-range overlap among represent the data, we used this model as a basic description for neighboring territories was low, averaging .% (range: .–%). between-sample comparisons and for comparison to density esti- The sub-adult male’s home range overlapped with that of another mates. All analyses were conducted using PRESENCE, version . male that occupied a dancing ground. Mean daily home ranges (MacKenzie et al. ). were approximately – ha and did not differ significantly among We estimated densities of Great Argus Pheasants using line- birds. Home-range sizes did not differ significantly by month (F  transect sampling methods (Burnham et al. , Buckland et al. ., df   and , P  .; Table ). ). Each month, from June  to January , three pairs of The distribution of radiolocations within home ranges of observers walked  transects in the Way Canguk research area adult males was concentrated around the birds’ dancing grounds. over a three-day period for a total of . km month−. Transects Kernel analysis showed that core areas (% of use) of male Great were walked each day by observer pairs spaced at -m intervals, Argus Pheasants were very small, at –% of total home-range beginning at  hours and ending at ~ hours. We recorded size. All males had multiple centers of activity. Three of the six the total length of transect walked and the number of clusters adult males centered their activities close to dancing grounds (consisting of one or more birds) detected. For each cluster, we that were located near the boundaries of the home range. Danc- noted the time of observation, detection cue (audio, visual), num- ing grounds of the other three adult males were in the middle of ber of individuals, sex when possible, distance to cluster, and angle the home range. These individuals moved throughout their home from transect to cluster. All surveys were evaluated for the possi- ranges more uniformly. bility of duplicate observations that might arise from more than Daily movements of male Great Argus Pheasants ranged be- – one observer recording a loud call. Duplicate records were elimi- tween  and , m (x   o  m; Table ) with significant nated. We calculated density estimates for each of five samples by differences among months (F  ., df   and , P  .). Daily – combining census data for a three-month period that centered on movements were significantly shorter in September (x   m) – the month of camera trapping in the study area using DISTANCE, than in October (x   m). Although there were differences in version . (Burnham et al. ; Buckland et al. , ). We the number of locations near dancing grounds, we found no sig- initially calculated detection probabilities for each sample sepa- nificant monthly differences in distances to dancing grounds rately, but, because of limited sample sizes, we pooled observa- (F  ., df   and , P  .). Adult male Great Argus Pheasants’ tions to obtain a single detection function and applied this to each mean distances to the dancing ground varied from  to  m – – sample to obtain final density estimates. We made separate den- (x   o .) and showed an increasing trend over time (x  – – Sept. sity calculations for males detected by calls and an overall density  m, xOct.   m, xNov.   m). based on combined visual and call cues. Correlation of food abundance and home range.—The index of plant food abundance (fruits, flowers, seeds, and fungi) was gener- RESULTS ally low and did not show any particular pattern or trend over the three months of the study. We found that home-range size was not Male home-range and movement patterns.—In , between July related to availability of plant food for the food types measured. and mid-September, we captured nine male Great Argus Pheasants Also, daily ranging patterns and distance moved from dancing during a total of  trap-hours of effort. There was no indication grounds were unrelated to availability of plant food. that the birds were injured by the snares, and no bird died as a re- Habitat characteristics of home range.—Home ranges sult of capture stress. One bird, however, was killed by an unknown and point locations were divided among macrohabitat types.

TABLE 1. Minimum convex polygon (MCP) monthly and three-month total home ranges (ha) of male Great Argus Pheasants during September– November 2001, in Bukit Barisan Selatan National Park, Sumatra, Indonesia.

Number Individual Age of locations September October November Total

AAdult1229.28.06.616.1 BAdult1127.49.47.012.6 C Adult 116 4.6 9.5 7.5 14.5 D Adult 113 2.9 4.9 1.8 7.6 E Adult 82 6.3 5.9 No data 8.3 G Adult 86 2.7 9.5 2.2 10.0 F Subadult 105 13.8 8.4 7.6 32.3 346 — WINARNI ET AL. — AUK, VOL. 126

TABLE 2. Means (o SE) of daily travel (m) of male Great Argus Pheasants during September–November 2001, in Bukit Barisan Selatan National Park, Sumatra. Daily tracking was conducted twice a month for each bird (an asterisk indicates that only one daily tracking was conducted).

Individual September October November

A 1,145.1 o 127.0 1,084.5 o 253.2 907.1 o 484.1 B 649.0 o 70.0 1,001.5 o 139.3 821.7 o 119.8 C 412.7 o 86.8 1,024.2 o 85.0 924.2 o 327.0 D 629.2 o 122.9 733.7 o 26.8 626.4 o 86.9 E 979.0 o 184.8 847.1 o 240.7 No data G578.4*1,042.0 o 422.1 595.3 o 5.5 F 815.8 o 24.3 1,087.4 o 170.1 1,073.4 o 25.6

Compositional analysis indicated the same rank order of habitats in the study area and in the MCP ranges. Home ranges were lo- cated primarily in habitat I, and home ranges contained more of FIG. 2. Density estimates with 95% confidence intervals for calling males habitat II and less of habitat IV than the study area. Only two home and all Great Argus Pheasants between July 1998 and November 2001. ranges contained more disturbed than undisturbed habitat. Over- all, however, there were no significant differences between avail- ability of habitats and occurrence of habitats in the MCP ranges. Canguk research area. Observations were truncated at  m. Although MCP ranges included all habitat types, radiolocations Great Argus Pheasants were detected  times during line- of male Great Argus Pheasants occurred primarily in undisturbed transect surveys, and % of these detections were of calling – – habitats I (x  %) and II (x  .%). Compositional analysis also males. We rarely distinguished sex during visual encounters and identified different rank order in radiolocations compared with could not analyze male and female density separately using vi- proportion to habitat available within the home range. Use of hab- sual encounters. We calculated detection probability for each of itat I was significantly higher than use of habitat III (T  ., df  , the five calling-male samples and found no significant differences P  .) or habitat IV (T  ., df  , P  .). Use of habitat I (range: .–.) among detection probabilities, so we combined was greater than use of habitat II, and use of habitat II was greater data sets to estimate the detection function by fitting a half-nor- than use of habitats III and IV, but these differences were not sig- mal distribution with cosine adjustment terms. The detection nificant. Comparison of percentage of use of the % MCP home probability (o SE) of calling males was . o ., and effective ranges with probabilistic kernel home ranges indicated that male strip width was . o . m. Densities of calling males increased Great Argus Pheasants were most likely to use habitat I and avoid from . km− to a maximum of . km− and then declined to habitat IV. Habitat IV is represented as % of habitat use in the % . km− in the final survey (Fig. ). Overall male density showed kernel and % of habitat use in the % kernel representing the an increasing trend over time, with a geometric mean increase of core range. By contrast, habitat I is represented as % and % of % year− between the first and last surveys. habitat use in the % kernel and % kernel, respectively. Use of We estimated the detection function for males and females habitats II and III appears to be similar to availability. by combining calling and visual observations and fitting a half- Habitat characteristics of display sites.—We classified  ac- normal distribution with cosine adjustment terms. Birds were tive dancing grounds to macrohabitat and measured microhab- detected with an average probability of . o ., and the ef- itat characteristics of the dancing grounds and  random sites fective strip width was . o . m.The total density estimate in- within home ranges of male Great Argus Pheasants. Most danc- creased from . km− to a maximum of . km− and declined ing grounds were located in undisturbed forest (habitat I, n  ; to . km− in the final survey. Total population increased at a habitat II, n  ; habitat III, n  ). The difference between the ob- rate similar to that of calling males, with a geometric mean of % served and expected distributions of dancing grounds among the year− between the first and last surveys. four habitat types was marginally significant (C  ., df  , Camera-trap surveys resulted in  independent photo- P  .). A forward-stepwise binary logistic regression of micro- graphs of single birds,  photographs of two birds, and three pho- habitat variables at the display sites and random points resulted tographs of three birds. Birds were photographed throughout the in a model that retained lianas (C  ., df  , P < .) and day (– hours) but were most active between  and leaf size (C  ., df  , P < .) as significant explanatory  hours (% of photographs). Great Argus Pheasants were de- variables. Display sites were characterized by having fewer lianas tected at least once at  locations in each survey in Way Canguk, nearby and smaller leaf sizes on adjacent trees, compared with for an average observed or naïve occupancy estimate (assuming random points. The model then correctly classified  of the  detection probability  ) of Great Argus Pheasant occurrence sites used to create the model. equal to % (range: –%) of camera locations. Occupancy es- – Abundance and distribution.—We used  line-transect sur- timates were % higher, on average (x  %, range: –%), and veys, grouped in three-month intervals centered on five camera- more consistent than the naïve estimates. Naïve estimates were trap surveys between August  and January  in the Way consistently lower, and in the April  sample fell below the % APRIL 2009 — GREAT ARGUS PHEASANT MOVEMENT AND ABUNDANCE — 347

were not accompanied by expansion into unoccupied habitat. Our results also indicate that populations of Great Argus Pheasant are subject to rapid changes in abundance over a short period. Habitat specificity and rapid population fluctuations may make Great Ar- gus Pheasants especially vulnerable to the effects of forest loss and degradation, as well to the increasing frequency of ENSO events. Movement patterns, display sites, and food abundance.—The average male home ranges in our study site, both monthly (. o . ha) and over three months (. o . ha), are consider- ably larger than home-range estimates collected over five months from peninsular Malaysia (.–. ha; Davison a). Core- area use averaged <% of the MCP in Sumatra and consistently covered larger areas than observed in Malaysia (.–. ha and .–. ha, respectively). Daily movements of Great Argus Pheas- ants were also larger in Sumatra than in Malaysia. In Sumatra, birds traveled an average of  o  m daily, whereas in Malay- sia, the maximum distance traveled in a day was  m and  m for each of two birds. There are some temporal and habitat differ- ences between the two studies. Whereas our study was conducted at the end of the presumed breeding season, Davison’s (a, b) spanned the pre-breeding and breeding seasons, as determined by maintenance of display sites and calling. The Malaysian site was hilly, the forest was dominated by the tree family Dipterocar- paceae, and both birds maintained their display site on the tops of ridges and moved about on the hillside below. In Way Canguk, the FIG . 3. Locations of cameras traps in the south study area of Way Canguk site was flat, and although Dipterocarpaceae was the most numer- in relation to the four dominant habitat types. Filled circles indicate loca- ous family in the canopy, this family did not dominate the forest. tions where Great Argus Pheasants were photographed at least once dur- Davison and Scriven () reported that display sites were spaced ing five surveys between 1998 and 2001. more widely on flat, lowland sites than on hilly sites and that den- sities of calling males were lower at flat sites. Both results imply confidence interval (CI) for estimated occupancy. As was observed either larger territories or a patchier distribution of territories on for males in the radiotelemetry study, Great Argus Pheasants oc- flat terrain in Malaysia. curred more often at cameras that were adjacent to or surrounded Movements and territory size of male Great Argus Pheas- by intact habitat (habitats I and II; C  ., df  , P  .; Fig. ), ants may be affected by food abundance (Davison a, Johnsgard and no individuals were photographed by cameras in highly dis- ) as well as by the need to stay close to a display site during the turbed forest. Only one photograph was obtained at a trap site that breeding season. We found that our index of plant-food abundance was not adjacent to intact habitat. was unrelated to male territory size or daily movement patterns. To test for possible habitat separation between males and fe- However, Great Argus Pheasants are omnivorous, and we did not males, we examined the distribution of males and females across measure leaf-litter invertebrates in the present study. We observed films at locations where they were photographed. For each sample that daily distance traveled increased over time, possibly coinciding period, camera points were scored as “only males photographed” with the end of the breeding season. We know very little about the (n  ), “only females photographed” (n  ), and “both sexes breeding season of Great Argus Pheasant (Madge and McGowan photographed in the same sample” (n  ). If the sexes were seg- , present study), however, except that nests have been found regated spatially in Way Canguk, we expected a preponderance in May, June, July, August, and October (Davison a, Johns- of camera points with one sex and few points with both sexes. We gard , M. F. Kinnaird pers. obs.). If display sites are a limited found no evidence that sexes were segregated among the camera- and defended resource, as they are in many lekking species (Davi- trap locations (C  ., df  , P  .). son b, Madge and McGowan ), there should be a tradeoff between foraging and defending a territory. Davison (a) sug- DISCUSSION gested that during the breeding season, possession of a display site and the amount of time spent there are critical to a male’s fitness. Our telemetry results indicate that male Great Argus Pheasants He also suggested that Great Argus Pheasants are able to mini- are highly territorial during the breeding season and that male mize their energy expenditure during low food availability by lim- movements appear to be constrained more by the need to attend iting movements and spending –% of the day resting (Davison and possibly defend the display site than by foraging needs. Our a). In Bukit Barisan Selatan National Park, birds were active results also show that Great Argus Pheasants, males and females, throughout the day, though % of activity occurred between  are habitat specialists, preferring intact forest with large trees and and  hours (O’Brien and Kinnaird ). open understory, and show no evidence of habitat separation. In- Forests dominated by Dipterocarpaceae are usually char- creases in the abundance of Great Argus Pheasants over . years acterized as fruit-poor during non-masting years (Leighton and 348 — WINARNI ET AL. — AUK, VOL. 126

Leighton , Wich and Van Schaik ). Fruit specialists in closely linked to time since disturbance, with old secondary for- Dipterocarpaceae-dominated forests must be able to move widely est preferred over young secondary forest. In the present study, over the landscape (Kinnaird and O’Brien ) to secure suffi- undisturbed forest with large trees was preferred; this forest type cient resources. Although Way Canguk is not dominated by Dip- has a less dense understory, preferred for display sites, and a well- terocarpaceae, only % of all tree species were in fruit (at all stages developed litter layer, which is critical as a foraging substrate (Da- of ripeness) in any month during the study. Given the small male vison a, b). territory size, constrained movements, and low plant-food index Abundance and distribution.—The estimated density of call- values, it is possible that males relied more heavily on invertebrates ing males at Way Canguk increased from . to a peak of . males as a food supply during this study period. Beebe () reported km− and then declined to . males km− over a .-year period that the Bornean subspecies inhabiting Dipterocarpaceae-domi- following a major ENSO drought and fire (Kinnaird and O’Brien nated forests ate primarily ants but also leaves, fruits, and seeds. ). At the same time, density of all Great Argus Pheasants in- Davison (a) analyzed droppings and reported that solitary creased from . birds km− to a peak of . birds km− and declined leaf-litter ants were the primary invertebrate food of the Malay– to . birds km−. Both data sets indicate a three-fold increase in Sumatra subspecies but that fruits were predominant in the diet. Great Argus Pheasants between  and . Three potential Ants and other leaf-litter invertebrates are usually exploited by problems regarding the density results should be considered. () systematically exploring small patches, a description that fits the Detection probabilities, rather than density, varies over time, and meandering foraging behavior of Great Argus Pheasants (Davi- use of pooled data obscures that relationship. () Sampling error son a). Davison (a) concluded that the most likely changes inflated results for September . And () overall densities were in the foraging patterns of Great Argus Pheasants would be re- underestimated because of sex-specific differences in detectabil- lated to changes in diet and changes in the speed and amount of ity between males and females. time spent searching for food. If plant food resources were, in fact, Detection probabilities ranged from . to . across sam- scarce during the present study, the increases in movements over ples, but these differences were not significant. Maximum and a restricted area we recorded might be explained by invertebrate minimum detection probabilities did not correspond to maxi- foraging. However, the short duration of our study and the lack of mum and minimum density estimates, and the correlation be- data on leaf-litter invertebrates preclude firm conclusions. tween detection probability and density was not significant Habitat preference.—Several lines of evidence point to strict (P  .). In , we recorded  active display sites and esti- habitat specificity in Great Argus Pheasants. First, male display mated  calling males (% CI: .–.) from the line-transect sites tend to be located in undisturbed forest with large trees (hab- data, which is consistent with results of independent surveys. Sam- itat I, P  .), and  of the  dancing grounds were found in pling error could explain the apparent outlier density recorded in the undisturbed habitats. Display-site characteristics included an- September . However, during larger camera-trap surveys of imal trails, presence of low branches, and low density of seedlings, the ,-km national park (O’Brien and Kinnaird ), we re- saplings, lianas, herbs, and gingers. Lianas, herbs, and gingers are corded a similar pattern of abundance between  and , characteristic of disturbed sites at Way Canguk. Nijman () with point abundance estimates tripling in BBSNP between  reported that of six active display sites located during surveys in and , followed by a decline. Borneo, five occurred in primary forest and one in old secondary The final problem of possible sex bias in detection probabili- forest. Davison (b) reported that  display sites were found in ties is more complex. If the sex ratio is skewed (often the case for primary and tall secondary forest. polygamous species) and there are sex-specific detection proba- Second, radiotelemetry suggests that home ranges of male bilities (males detected more often), density estimates based on Great Argus Pheasants were centered on undisturbed forest, pooled samples might be biased. Davison (a) suggested that habitat I in particular. Although MCP home ranges incorporated both sexes of Great Argus Pheasant maintain exclusive feeding habitats I–IV in similar order to their representation at the site, territories during and immediately after the breeding season. He radio locations placed male Great Argus Pheasants firmly in habi- based this conclusion on observations of solitary foraging, dis- tat I (% of locations) and habitat II (% of locations). Habitat persion of invertebrate prey, and optimal-foraging considerations I was used significantly more than all other habitats and consti- (Pyke et al. ). If Davison’s (a) hypothesis is supported, and tuted % of the habitat in the core area. Other habitats were ei- discrete female territories overlap discrete male territories, we ther avoided (habitat IV) or used in proportion to occurrence. would expect a sex ratio similar to the ratio of respective territory Third, camera-trap surveys confirmed that male and fe- sizes, because males and females would space themselves across male Great Argus Pheasants were associated with undisturbed the landscape in discrete territory patches. In addition, point sam- habitat, with no evidence of spatial segregation between sexes. ples should record a : sex ratio, given that, on average, any point In Way Canguk, individuals were photographed at a camera site would fall in just one male and one female territory. We estimated surrounded by disturbed forest only once over a .-year period the sex ratio of Great Argus Pheasants from camera-trap data for (, camera-trap days of effort). As the Great Argus Pheasant Way Canguk and BBSNP in two ways. First, we counted the num- population increased in Way Canguk, the estimated proportion of ber of male and female Great Argus Pheasants in  independent occupied habitat did not increase, which indicates that the species photographs (T. O’Brien unpubl. data), excluding photographs of used only a subset of the total study area. unsexed birds. If we assume no habitat separation and that males Most authors agree that Great Argus Pheasants prefer pri- and females pass through the camera at the same rate, the ratio mary forest and old secondary forest (Davison a, Nijman of birds in photographs should reflect the sex ratio. We estimated , Johnsgard ). Use of secondary forest appears to be a sex ratio of . males to . females. Next we assumed that a APRIL 2009 — GREAT ARGUS PHEASANT MOVEMENT AND ABUNDANCE — 349

camera-trap point was within, at most, one male and one female Winarni et al. () found Great Argus Pheasants in only  of  territory. The ratio of males to females appearing at least once at forest fragments (ranging in size from . to  km), with vary- a camera-trap point was .:.. These sex ratios support Davi- ing degrees of isolation. All forest areas  km contained Great son’s (a) hypothesis of female and male territories. Finally, if Argus Pheasants. As deforestation continues across Peninsular there are sex differences in detection probabilities, we would ex- Malaysia, Sumatra, and Borneo, large undisturbed tracts of low- pect the sex ratio of male density to female density to differ from land forest will become a limiting resource for Great Argus Pheas- :. We can derive the sex ratio for each line-transect survey as ants and other species that depend on high-quality lowland forest, the ratio of the male density estimate to the total density estimate such as peacock pheasants (Polyplectron spp.), Wrinkled Horn- minus the male density estimate. In line-transect surveys  and , bill (Rhyticeros corrugatus), and Black Hornbill (Anthracoceros estimated sex ratios are :; in surveys  and , sex ratios are male- malayanus). Strict habitat-specificity reduces the likelihood that biased (males:females  .:. and .:., respectively); but in Great Argus Pheasants will recolonize empty forest fragments. survey , the sex ratio is female-biased (.:.). We conclude that there is no consistent evidence of sex bias in detection probabili- ACKNOWLEDGMENTS ties using line-transect sampling. Density estimates indicate that populations of Great Argus This work was funded by the Wildlife Conservation Society, by Pheasant may fluctuate considerably over a relatively short time- Warnell School of Forest Resources, by the Office of the President scale and that timing of surveys can have a large effect on the re- of the University of Georgia, and by the Georgia Game Bird Breed- sults observed. We suspect that the populations in the study area ers Association. We thank the Bukit Barisan Selatan National Park and BBSNP were depressed following one the most severe ENSO [BBSNP]–Department of Forestry for permitting us to work in BB- droughts of the th century (Kinnaird and O’Brien ), and SNP. C. Sheppard provided helpful comments and support during the steep increase between  and  may represent a popu- testing of radio necklaces on Great Argus Pheasants at the Bronx lation recovery. Alternatively, density differences may represent Zoo. We also thank H. T. Wibisono, M. Iqbal, A. Nurcahyo, M. Si- annual fluctuations in population size resulting from birth, re- naga, and our field assistants, Marwoto, D. Suyadi, Waryono, Aris, cruitment and mortality. Data on reproductive patterns in Great and Teguh, for their assistance with the data collection. The manu- Argus Pheasants are scarce: we observed only a single nest in Au- script benefited from the comments of two anonymous reviewers. gust , and chicks in November , January , and No- vember . This suggests that most recruitment occurs later in LITERATURE CITED the year. Clutch size is two eggs, but multiple clutches in a year are possible (Johnsgard ). Aebischer, N. J., P. A. Robertson, and R. E. Kenward. . The density estimates at Way Canguk are similar to estimates Compositional analysis of habitat use from animal radio-tracking for primary and old secondary forests from other published stud- data. Ecology :–. ies (Davison a, Davison and Scriven , Nijman ). These Beebe, W. . Pheasant: Their Lives and Homes, vol. . New York studies, however, reported consistently higher densities on hilly Zoological Society, New York. slopes (males: .–. km−, total: .–. km−) than on flat ar- Buckland, S. T., D. R. Anderson, K. P. Burnham, and J. L. eas (males: .–. km−, total: .–. km−). Wells () Laake. . Distance Sampling: Estimating Abundance of Bio- considered Great Argus Pheasants slope specialists, though with logical Populations. Chapman & Hall, New York. little supporting data. Our data suggest that Great Argus Pheas- Buckland, S. T., D. R. Anderson, K. P. Burnham, J. L. Laake, ants may reach high densities in flat forest, so long as the forest is D. L. Borchers, and L. Thomas. . Introduction to Distance not inundated or heavily disturbed. Such forest conditions are be- Sampling. Oxford University Press, Oxford, United Kingdom. coming increasingly rare in Southeast Asia (Kinnaird et al. , Burnham, K. P., D. R. Anderson, and J. L. Laake. . Estima- Kinnaird and O’Brien , Sodhi et al. ). tion of density from line transect sampling of biological popula- Although density in Way Canguk fluctuated considerably be- tions. Wildlife Monographs, no. . tween surveys, the overall trend in numbers of Great Argus Pheas- Davison, G. W. H. a. Diet and dispersion of the Great Argus ants was an increase over time. The proportion of occupied habitat Argusianus argus. Ibis :–. in Way Canguk remained relatively constant and approximated Davison, G. W. H. b. Sexual selection and the mating system the proportion of undisturbed habitat with large trees. Way Can- of Argusianus argus (Aves: ). Biological Journal of the guk did not experience fires or forest loss during the study period, Linnaean Society :–. so proportions of habitat remained relatively unchanged. As their Davison, G. W. H., and K. Scriven. . Recent pheasant surveys abundance increased, Great Argus Pheasants did not expand their in peninsular Malaysia. Pages – in Proceedings of the nd foraging areas or breeding territories to disturbed habitats. These International Symposium on the Pheasants in Asia (C. Savage and results agree well with the core-area results from the telemetry M. W. Ridley, Eds.). World Pheasant Association, Reading, United study of males that show overwhelming use of undisturbed forest, Kingdom. particularly habitat I. Avoidance of disturbed forest may explain Delacour, J. . The Pheasants of the World. World Pheasant why Thiollay () recorded no Great Argus Pheasants in tradi- Association and Spur Publications, Hindshead, United King- tional rubber, dammar (resin), or durian agroforests adjacent to dom. primary forest in BBSNP. Fuller, R. A., and P. J. Garson. . Pheasants: Status survey and As sedentary habitat-specialists, Great Argus Pheasants conservation action plan –. WPA/BirdLife/SSC Pheas- are not likely to move long distances across unsuitable habitat. ant Specialist Group, IUCN, Gland, Switzerland. 350 — WINARNI ET AL. — AUK, VOL. 126

Holmes, D. A. . The predicted of lowland forests O’Brien, T. G., and M. F. Kinnaird. . Birds and mammals in Indonesia. Pages – in Terrestrial Ecoregions of the Indo- of the Bukit Barisan Selatan National Park, Sumatra, Indonesia. Pacific: A Conservation Assessment (E. Wickramanayake, E. Oryx :–. Dinerstein, C. J. Loucks, D. M. Olson, J. Morrison, J. Lamoreux, O’Brien, T. G., and M. F. Kinnaird. . A picture is worth a M. McKnight, and P. Hedao, Eds.). Island Press, Washington, thousand words: The application of camera trapping to the study D.C. of birds. Bird Conservation International :S–S. Hooge, P. N., and B. Eichenlaub. . Animal Movement exten- O’Brien, T. G., M. F. Kinnaird, and H. T. Wibisono. . Crouch- sion to ARCVIEW, version .. U.S. Geological Survey, Alaska Bio- ing tigers, hidden prey: Sumatran tiger and prey populations in a logical Science Center, Anchorage. [Online.] Available at www. tropical forest landscape. Animal Conservation :–. absc.usgs.gov/glba/gistools/animal_mvmt.htm. Ott, P., and F. Hovey. . BYCOMP.SAS. Clearinghouse for Johnsgard, P. A. . The Pheasants of the World: Biology and Ecology Software. Illinois Natural History Survey, Champaign, Natural History, nd ed. Smithsonian Institution Press, Wash- Illinois. ington, D.C. Pyke, G. H., H. R. Pulliam, and E. L. Charnov. . Optimal Kinnaird, M. F., and T. G. O’Brien. . Ecological effects of foraging: A selective review of theory and tests. Quarterly Review wildfire on lowland rainforest in Sumatra. Conservation Biology of Biology :–. :–. Smythies, B. E. . The Birds of Borneo, rd ed. Sabah Society and Kinnaird, M. F., and T. G. O’Brien. . The Ecology and Con- Malayan Nature Society, Kuala Lumpur. servation of Asian Hornbills: Farmers of the Forest. University of Sodhi, N. S., M. Rose, C. Posa, T. M. Lee, and I. G. Warkentin. Chicago Press, Chicago. . Effects of disturbance or loss of tropical rainforest on birds. Kinnaird, M. F., E. W. Sanderson, T. G. O’Brien, H. T. Wibi- Auk :–. sono, and G. Woolmer. . Deforestation trends in a tropi- Sunarto. . Survival of plant communities and dynamics of for- cal landscape and implications for endangered large mammals. est ecosystems in response to fire: A long-term study from Sumatra. Conservation Biology :–. MSc. thesis, University of East Anglia, Norwich, United Kingdom. Leighton, M., and D. R. Leighton. . Vertebrate responses to Thiollay, J.-M. . The role of traditional agroforests in the con- fruiting seasonality within a Bornean rainforest. Pages – servation of rain forest bird diversity in Sumatra. Conservation in Tropical Rainforest Ecology and Management (S. L. Sutton, Biology :–. T. C. Whitmore, and A. C. Chadwick, Eds.). Blackwell Scientific, Wells, D. R. . The forest avifauna of western Malesia and its Oxford, United Kingdom. conservation. ICBP Technical Publications :–. MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, White, G. C., and R. A. Garrott. . Analysis of Wildlife J. A. Royle, and C. A. Langtimm. . Estimating site occu- Radio-tracking Data. Academic Press, San Diego, California. pancy rates when detection probabilities are less than one. Ecol- Whitten, T., S. J. Damanik, J. Anwar, and N. Hisyam. . ogy :–. Ecology of Sumatra, nd ed. Periplus Editions, Singapore. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. P. Pollock, L. L. Wich, S. A., and C. P. Van Schaik. . The impact of El Niño Bailey, and J. E. Hines. . Occupancy Estimation and Mod- on mast fruiting in Sumatra and elsewhere in Malesia. Journal of eling: Inferring Patterns and Dynamics of Species Occurrence. Tropical Ecology :–. Academic Press, New York. Winarni, N. L. . Avifauna dan mamalia. In Laporan Hasil Pene- MacKinnon, J., and K. Phillipps. . A Field Guide to the Birds litian TNBBS WCS-IP. Wildlife Conservation Society-Indonesia of Borneo, Sumatra, Java, and Bali. Oxford University Press, New Program, Bogor. York. Winarni, N. L. . Distribution and movement patterns of Great Madge, S., and P. McGowan. . Pheasants, Partridges and Argus Pheasant (Argusianus argus) in Bukit Barisan Selatan, Grouse: A Guide to the Pheasants, Quails, Grouse, Guineafowl, Sumatra, Indonesia. M.Sc. thesis, University of Georgia, Athens. Buttonquails, and Sandgrouse of the World. Princeton University Winarni, N. L., A. Nurcahyo, Y. Hadiprakarsa, and M. Press, Princeton, New Jersey. Iqbal. . Effects of forest patch size on in south- McGowan, P., and M. Gillman. . Assessment of the conser- ern Sumatra, Indonesia. Pages – in Galliformes . Pro- vation status of partridges and pheasants in South East Asia. Bio- ceedings of the rd International Galliformes Symposium (R. A. diversity and Conservation :–. Fuller and S. J. Browne, Eds.). World Pheasant Association, Ford- Nijman, V. . Habitat preference of Great Argus Pheasant (Argu- ingbridge, United Kingdom. sianus argus) in Kayan Mentarang National Park, East Kaliman- tan, Indonesia. Journal of Ornithology :–. Associate Editor: J. H. Vega Rivera